The Doherty power amplifier was first proposed in 1936 by W. H. Doherty his article Doherty, W. H., “A New High Efficiency Power Amplifier for Modulated Waves,” Proceedings of Institute of Radio Engineers, pp. 1163-1182, September 1936. While the Doherty power amplifier initially had success using vacuum tube amplifiers, the Doherty power amplifier essentially ceased to exist starting in the 1960s and 1970s with the introduction of solid state transistors. In particular, the Doherty amplifier requires precise matching between the main and auxiliary amplifiers. However, due to tolerances in solid state transistors, the required matching between the main and auxiliary amplifiers could not be achieved, which in turn caused Doherty power amplifiers implemented using solid state transistors to cease to operate as intended.
With the advent of digital signal processing and, in particular, digital predistortion, Doherty power amplifiers have recently reemerged as a power amplifier of choice, especially for base stations in cellular communications networks. Specifically, digital signal processing, such as digital predistortion, can now be utilized to compensate for mismatches between the main and auxiliary amplifiers. As a result, Doherty power amplifiers can now be implemented using solid state transistors. Due to the significant improvement in the efficiency of the Doherty power amplifier as compared to other power amplifier architectures, Doherty power amplifiers are now the preferred amplifiers for modern cellular communications standards, which have a high Peak-to-Average Power Ratio (PAPR).
One issue with existing Doherty power amplifier architectures is that they have a limited back-off power level in which the efficiency of the Doherty power amplifier is maintained. Typically, the back-off power level is 6-8 decibels (dB) such that the Doherty power amplifier is efficient for power levels from its maximum power level to 6-8 dB below its maximum power level. However, existing Doherty power amplifiers have poor efficiency when operated considerably below their limited back-off power level.
The limited back-off power level in existing Doherty power amplifiers is problematic because the average power of future base stations is expected to fluctuate in a range of up to 20 dB. This will lead to inefficient power amplification during most of the transmission time of the base station if an existing Doherty power amplifier, which has a back-off power level of 6-8 dB, is used.
In the paper Darraji, R. et al., “Digital Doherty Amplifier With Enhanced Efficiency and Extended Range,” IEEE Transactions on Microwave Theory and Techniques, Vol. 59, No. 11, pp. 2898-2909, November 2011, the authors proposed a Doherty power amplifier that utilizes a digitally controlled dynamic input power distribution scheme that extends the back-off power level range in which the Doherty power amplifier maintains high efficiency. While this technique was initially sought to mitigate the imperfect load modulation mechanism due to the difference in class of operations of the carrier and peaking amplifiers, the consequential early saturation of the carrier amplifier resulted in a theoretical extension of the efficiency range by 3 dB at best. This efficiency range extension is insufficient to tackle the significant efficiency deterioration for an important variation of the input average power level.
In the paper Gustafsson, D. et al., “A Modified Doherty Power Amplifier with Extended Bandwidth and Reconfigurable Efficiency,” IEEE Transactions on Microwave Theory and Techniques, Vol. 61, No. 1, pp. 533-542, January 2013, the authors proposed a Doherty power amplifier in which the power back-off level can be reconfigured by changing the drain-source bias of the main amplifier. However, this technique suffers from three performance issues and an architectural issue. Specifically, regarding the performance issues, this technique suffers from: (1) a low power utilization factor when reconfigured for high PAPR levels, (2) low efficiency at reduced average input power levels, and (3) strong dynamic nonlinearity that can seriously compromise the ability to meet the standards specifications. As for the architectural issue, this technique requires two separate RF inputs so that separate baseband signal processing functions can be applied to each path to ensure correct Doherty operation at different frequencies. The additional RF input results in additional complexity from an architecture standpoint.
In light of the discussion above, there is a need for a new Doherty power amplifier having an extended back-off power level while also maintaining efficiency.